Covalent functionalization of graphene oxide with proteins

(1)

Covalent functionalization of graphene oxide with proteins

M.Sc. thesis

University of Jyväskylä Department of Chemistry 3.6.2021

Elsa Korhonen

(2)

(3)

Abstract

The literature part of the master’s thesis focuses on introducing potential functionalization methods for the covalent attachment of proteins on graphene oxide (GO) surfaces. The experimental part aimed at the covalent immobilization of horseradish peroxidase (HRP) on the laser-oxidized graphene-based microchip. Based on the preliminary studies with graphene oxide flakes, both crosslinker systems, glutaraldehyde (GA) and APTES-glutaraldehyde, resulted in a similar outcome of the HRP immobilization. The thermogravimetric analysis and FTIR spectroscopic results suggested the formation of covalent bonds between the components of the GO-crosslinker-HRP systems.

The GA-based crosslinking method was chosen for the protein immobilization studies with the graphene-based microchip. From the atomic force microscopy (AFM) images of the microchip after the treatment with GA and HRP solutions (in PBS buffer), line-shaped structures and elevated dots could be observed, assigned to be GA crosslinkers and HRP protein molecules, respectively. The Raman spectra of the oxidized areas showed shifting of the D, G, and 2D bands towards lower Raman shifts after HRP immobilization, which agree with the former results, indicating a successful immobilization of HRP. Also, the D band areas of the oxidized regions increased after HRP immobilization, suggesting an increased number of defects in the graphene lattice. Based on the AFM and Raman spectroscopy results from the experiments with the chip, the covalent attachment of HRP to the chip’s surface via GA crosslinker could not be fully proved.

(4)

Tiivistelmä

Pro gradu -tutkielman kirjallisuusosassa perehdyttiin menetelmiin, jotka voisivat soveltua proteiinien kiinnittämiseen grafeenioksidin pintaan kovalenttisen sitoutumisen kautta.

Kokeellisessa osassa oli tavoitteena kiinnittää linkkerimolekyylien avulla piparjuuriperoksidaasi (HRP) -entsyymimolekyylejä laserilla hapetetun grafeenin pintaan.

Alustavissa kokeissa testattiin kahta menetelmää HRP:n kiinnittämiseksi grafeenioksidihiutaleisiin. Menetelmät erosivat käytettyjen linkkerimolekyylien suhteen.

Ensimmäisessä menetelmässä käytettiin glutaraldehydiä ja toisessa APTES:sta ja glutaraldehydistä muodostuvaa linkkeriä. Termogravimetrinen analyysi ja FTIR-mittausten tulokset viittasivat kovalenttisten sidosten muodostumiseen grafeenioksidin ja linkkerien sekä linkkerien ja HRP:n välille. Glutaraldehydiin perustuva menetelmä valittiin varsinaiseen kokeeseen, jossa käytettiin laserilla käsiteltyä grafeenipohjaista mikrosirua. Mikrosirusta otetuista AFM-kuvista havaittiin kohonneita viivamaisia ja pistemäisiä rakenteita funktionalisoinnin jälkeen, joiden pääteltiin olevan glutaraldehydiketjuja ja HRP- proteiinimolekyylejä. Mikrosirun hapetetuista alueista mitatut Raman-spketrit viittasivat muutoksiin hapetetun grafeenin elektronirakenteessa: D-, G- ja 2D-vöiden Raman-siirtymät pienenenivät ja D-vyön pinta-ala kasvoi funktionalisoinnin jälkeen. Muutosten pääteltiin johtuvan grafeenioksidin ja linkkerien tai HRP:n välisistä vuorovaikutuksista. Tulosten perusteella ei kuitenkaan voitu varmistaa HRP:n kovalenttista kiinnittymistä mikrosirun grafeenioksidipintaan glutaraldehydin välityksellä.

(5)

Preface

The literature part of the master’s thesis was started in June 2020 and was completed in May 2021. The experimental part was carried out at the Nanoscience Center at the University of Jyväskylä from August 2020 to December 2020. The topic was defined to concentrate on covalent protein immobilization methods, which could be used for the preparation of functional materials for biomedical applications. I would like to thank Professor Maija Nissinen for supervising my master’s thesis and Doctoral Researcher Johanna Schirmer for supervising the experimental work. I would also like to thank Dr Manu Lahtinen for carrying out thermogravimetric analysis from the graphene oxide flake samples.

(6)

Abbreviations

6-(azidomethyl)-2-pyridinecarbaldehyde atomic force microscopy

3-aminopropyltriethoxysilane 3-aminopropyltrimethoxysilane bovine serum albumin

chymotrypsin

chemically reduced graphene oxide

copper-catalyzed azide-alkyne cycloaddition differential scanning calorimetry

1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide extracellular matrix

expressed protein ligation Fourier transform infrared glutaraldehyde

graphene oxide

horseradish peroxidase lactoperoxidase

methylene blue

meta-chloroperoxybenzoic acid azide

N-hydroxysuccinimide oxalate oxidase

phosphate-buffered saline

poly(3,4-ethylene dioxythiophene) 6AMPC

AFM APTES APTMS BSA ChT CRGO CuAAC DSC

EDC/EDAC ECM EPL

FTIR GA GO HRP LPO MB mCPBA N3

NHS OxOx PBS PEDOT

(9)

PEG pI RGD TEM TGA

polyethylene glycol isoelectric point

peptide of arginine, glycine and aspartic acid transmission electron microscopy

thermogravimetric analysis

(10)

LITERATURE PART

1 Introduction

Graphene oxide (GO) is an attractive material for biomedical applications over other carbon- based materials, owing to its better biocompatibility and ability to interact with biological cells and tissues. GO’s good biocompatibility is based on its small size (one atom thick layer), large surface area, hydrophilicity, mechanical properties resembling the mechanical behavior of biological tissues, and the ease of functionalization with molecules, such as biomolecules.

Therefore, functionalized GO materials have numerous potential applications in biomedical fields, such as in drug delivery, biosensing, bioimaging, cancer therapy, tissue engineering, and regenerative medicine.^1,2

Biomaterials used in neural tissue engineering and regenerative medicine must mimic the natural environment of neurons, the extracellular matrix (ECM). Neurons need specific physical and chemical cues for their normal growth and biological functions. GO’s electrical conductivity and oxygen-containing groups for functionalization with biological molecules make it advantageous material for nerve regeneration and neural interfacing applications.^3,4 Conductivity of GO improves neural cell-cell interactions, enhancing growth and regeneration of nervous tissue. In addition, the structural flexibility of GO enhances its ability to interact with cells and biomolecules, which further improves its biocompatibility.⁵

Although GO is a biocompatible material, it also has challenges in biomedical applications, such as cytotoxicity, biodistribution (non-biodegradability), and inflammatory responses.

These problems can be solved by modifying GO to resemble the natural microenvironment of neurons by functionalizing it with biomolecules, such as ECM proteins. By functionalizing GO with ECM proteins, a three-dimensional substrate for the attachment and growth of neurons can be achieved. Also, by attaching cell adhesive molecules on the GO surfaces, biocompatibility towards neural cells can be enhanced.^3,5 Studies related to covalent GO functionalization with ECM proteins are scarce but utilization of other proteins, such as enzymes, have been widely published.

(11)

This master’s thesis focuses on introducing covalent functionalization protocols of GO, which could be applicable for attaching proteins (compatible with nervous tissue) on the laser- oxidized graphene-based microchip. The literature part starts by introducing chemical properties and preparation methods of GO, and the most common methods used for the characterization of functionalized GO materials. Then, weak and covalent GO-protein interactions are presented with highlighting proteins’ chemical moieties participating in those interactions. Chapter 4 focuses on covalent protein immobilization methods on GO and their preceding chemical modification methods. The literature part ends with examples of ECM protein immobilization on GO. The experimental part aimed at the covalent immobilization of horseradish peroxidase (HRP) on the laser-oxidized graphene-based microchip using glutaraldehyde as a crosslinker. Preliminary experiments with GO flakes using two different crosslinker systems are also reported.

(12)

2 Graphene oxide

GO is a graphene derivative with a honeycomb-like plane of sp² hybridized carbons. Graphene consists of an aromatic delocalized π-system network, making it a hydrophobic and inert material. Pristine graphene sheets have poor dispersibility in organic solvents, and they aggregate easily in aqueous solutions. GO is an oxidized form of graphene, with many advantages compared with pristine graphene. Because of oxygen-containing functional groups, GO has versatile opportunities for covalent functionalization, and it is well dispersible in water and polar organic solvents. Homogenous dispersal of graphene derivatives in aqueous solutions is crucial in the preparation and some applications of functional nanomaterials.^6,7

2.1 Chemical structure

So far, any uncontroversial model to define the exact structure of GO has not been successfully developed. The most significant reasons for it are the nonstoichiometric atomic composition of GO and the lack of proper characterization methods.⁶ Also, the number and chemical nature of functional groups are highly dependent on the preparation method and starting material used.^8,9 It is commonly accepted that GO monolayers contain mainly OH and epoxy groups on their basal plane and COOH groups on the edges,⁶ but the real structure is more complicated. Also, attached OH groups can disrupt an otherwise flat surface morphology of GO sheets.¹⁰

Some proposed structural models for graphite oxide and GO are presented in this chapter.

Chemically graphite oxide and GO are quite similar, but their structures differ. Graphite oxide is a stacked, multilayered structure of GO sheets, whereas GO exist as monolayers or few- layered stacks. GO can be produced from graphite oxide through exfoliation by sonication or intensive stirring. Especially sonication can cause damage to GO sheets leading to significant differences in their sizes.⁶

(13)

The most cited and well-known model describing the graphite oxide structure is the Lerf- Klinowski model (1998).^10,11 The Lerf-Klinowski model classifies the structure of graphite oxide into two different regions: non-oxidized aromatic regions of sp² hybridized carbons and highly oxidized regions where sp³ hybridized carbons dominate. The relative sizes of these regions are determined by the oxidation degree of graphite oxide and the random distribution of the regions.¹¹ The highly oxidized regions of the graphite oxide’s basal plane contain OH and epoxy groups, whereas the edges of GO layers have OH and COOH groups (Figure 1). Lerf and Klinowski presented two structural models of graphite oxide, with and without COOH groups.^10,11 This model has been criticized because Lerf and Klinowski could not indicate the presence of COOH groups by their nuclear magnetic resonance (NMR) spectroscopic studies, probably owing to a low number of COOH groups.^11,12

Figure 1. Two suggested structures of graphite oxide based on the Lerf-Klinowski model, with and without COOH groups. Reprinted with permission from⁶, Copyright 1998, American Chemical Society (top). Reprinted from⁵, Copyright 1998, with permission from Elsevier (bottom).

(14)

The Lerf-Klinowski model is based on NMR spectroscopic studies,¹¹ whereas the earliest structural models of graphite oxide were based on elemental analysis, chemical reactions, and X-ray diffraction (XRD) methods.⁶ The first structural model of graphite oxide was presented in 1939 by Hofmann and coworkers.¹³ Based on the model, graphite oxide consisted of repeating units of 1,2-epoxides on the sp² hybridized graphene plane. Six years later, in 1946, Ruess¹⁴ proposed the model where graphite oxide consisted of the sp³ hybridized carbon plane containing 1,3-epoxides and OH groups. In 1969, Scholz and Boehm¹⁵ suggested the model in which graphite oxide contained hydroxyl and ketone groups but not epoxy groups.

Dekaný¹⁶ proposed a structural model of graphite oxide in 2006 using widely different techniques (NMR, elemental analysis, XRD, FTIR, transmission electron microscopy (TEM), electron spin resonance, and X-ray photoelectron spectroscopy). In Dekaný’s model, graphite oxide consisted of two distinct regions forming a corrugated carbon network: trans-linked cyclohexane chairs (1,3-epoxides and tertiary alcohols) and hexagon ribbons (cyclic ketones and quinones) (Figure 2).¹⁶

In addition to highly oxidized and non-oxidized regions of GO layers (the Lerf-Klinowski model), there may also be defected sites in graphene lattice. Defects can form due to harsh reaction conditions or overoxidation because of evolving CO and CO2 gases. Erickson et al.¹⁷

(15)

have imaged GO structure with TEM, which was oxidized from graphite by modified Hummers’

method¹⁸. The study proved that GO contains three different structural regions: highly oxidized areas (82 %), graphitic regions (16 %), and hole defects (2 %).¹⁷

Although structures of graphite oxide and GO have been studied extensively, new structural models are still presented. In 2019, Filiz et al.¹² proposed a GO structure based on NMR and FTIR studies (Figure 3), which supports the Lerf-Klinowski model, and has some similarities with Dekaný’s model. In the proposed structure, GO also contain lactone, ketone, and quinone groups in addition to hydroxyl, carboxylic acid, and epoxy groups. Also, some defects are included, which may form during graphite oxidation. Edges of the defects have mainly the same functionalities as the edges of GO sheets.¹²

2.2 Preparation methods

Graphite oxide is mainly produced from natural graphite by chemical oxidation. GO monolayers can be exfoliated from graphite oxide using sonication in organic solvents or water.

A commonly used method to produce graphite oxide from graphite is Hummers’ method (1958)¹⁸, in which graphite is oxidized by a mixture of sodium nitrate (NaNO3), potassium

(16)

permanganate (KMnO4), and concentrated sulfuric acid (H2SO4). A problem in this method is the formation of toxic gases, NO2 and N2O4,and a low yield.¹⁸ Therefore, several modifications or improvements of Hummers’ method have been presented. Better yields or more practical procedures have been achieved by replacing some reagents or increasing their amount. For example, by excluding NaNO3, increasing the amount of KMnO4 and adding phosphoric acid (H3PO4) can be prevented the formation of toxic nitrogen gases.¹⁹

Another example is Yu et al.’s²⁰ improved Hummers’ method with NaNO3-free protocol (Figure 4). The method reduces the amounts of the reagents (KMnO4 and H2SO4) needed in the synthesis by using K2FeO4 instead. In the first step (I), the intercalation compound of H2SO4

moleculesand graphite forms, which starts to turn into graphite oxide by K2FeO4/ KMnO4 pre- oxidation. In a deep oxidation step (II), more KMnO4 is added, and the intercalation compound turns entirely into oxidized graphite. In the presence of water (step III), hydrolysis and exfoliation of graphite oxide take place, leading to separate GO monolayers. The main steps of this protocol are similar to the original Hummers’ method except for the addition of KMnO4 in two steps and the use of K2FeO4. Also, this method is faster and gives better yields.²⁰

Figure 4. A scheme of the improved Hummers’ method to prepare GO monolayers, illustrating exfoliation of monolayers from oxidized graphite. In the article, the amounts and places of oxidants (K2FeO4 and KMnO4) are described other way round in the text than in the

http://creativecommons.org/licenses/by/4.0/.

(17)

Pumera and coworkers⁹ have studied the effects of reagents used in GO’s preparation for its chemical composition. They observed that permanganate-based oxidants result in a higher number of oxygen groups and a higher proportion of carbonyl and carboxyl groups than chlorate-based oxidants (e.g., Staudenmaier’s and Hofmann’s methods). Also, they found out that the strength of nitric acid (concentrated vs. > 90 %) used in the oxidation of graphite by chlorate-based oxidant did not significantly affect the GO’s oxidation degree.⁹ Hence, selecting the oxidant for graphite oxidation can affect the relative amounts of oxygen groups of the prepared GO.

Instead of chemical oxidation, graphene can be oxidized by the laser-induced two-photon oxidation.^21,22 The method enables tuning of oxidation levels of monolayer graphene by controlling laser pulse energies and irradiation times. Also, shapes of oxidized areas can be tuned by a directed laser beam with low pulse energies (~ 10 µW). Oxidation starts from point- like areas, which eventually combine into larger oxidized areas. In addition, oxidation probability is five times higher near the already oxidized areas than in pristine graphene.^21,22 Chemical composition of laser oxidized graphene resembles chemically prepared GO, but relative numbers of hydroxyl and epoxy groups differ: 40 % of carbons are involved in C-OH bonds and 25 % in C-O-C bonds, whereas chemically oxidized graphene usually have more epoxy groups. Additionally, laser oxidized graphene has a more ordered structure than chemically prepared.²³

2.3 Reactivity

Due to oxygen-containing functional groups, GO is more reactive towards covalent reactions than pristine graphene.²⁴ Also, GO’s surface is negatively charged in water due to partial deprotonation of OH and COOH groups, enabling electrostatic interactions with proteins.^25,26 GO can be functionalized covalently via its COOH, OH, and epoxy groups. The edges of GO are usually the most reactive regions towards covalent reactions because there are more space and, therefore, versatile opportunities to different bond angles.²⁷

(18)

COOH groups must be activated before functionalization, which can be done by carbodiimide formation (e.g., with EDAC) or treatment with thionyl chloride (SOCl2). The activated COOH groups can react with amines and alcohols, forming amide and ester linkages through carbodiimide coupling or condensation reaction. GO functionalization via its OH groups can also be done through carbodiimide coupling with a carboxyl group-containing reagent.

Additionally, OH groups can react with trialkoxysilanes and alkyltrichlorosilanes, forming silane linkages. GO functionalization via epoxy groups is mainly done through a ring-opening reaction using a nucleophilic reagent, usually an aromatic, aliphatic, or polymeric amine, or some carbanion.²⁴

GO flakes interact with each other via non-covalent interactions, mainly through π-π stacking or H-bonding. OH and epoxy groups on the basal plane induce stacking of GO platelets through interlayer H-bonding (Figure 5a). Also, intralayer H-bonding can occur. When GO flakes are in a water environment, the interlayer spacing of GO flakes increases due to intercalated water molecules via H-bonding (Figure 5b and 5c). The number of H-bonds between GO’s functional groups and water molecules increases as water content increases, enhancing the stacking of GO flakes. However, the high water content in the interlayer space of GO flakes eventually results in the degradation of GO interlayer bonds, reducing the stacking of GO flakes.²⁸

Figure 5. a) Inter- and intralayer H-bonding of functional groups of GO flakes. b) H-bonding between water molecules in GO’s interlayer space c) and between GO and water molecules.

(19)

2.4 Characterization methods of functionalized GO materials

Many methods are needed to confirm the success of the covalent functionalization of GO and the presence of protein molecules on GO surfaces. Shen et al.’s²⁹ results of bovine serum albumin (BSA) functionalized GO nanosheets demonstrate the importance of the wide scope of methods to analyze the covalent protein immobilization. Protein molecules were covalently attached to GO via a diimide-activated amidation reaction, which is described in more detail in Chapter 4.1.²⁹

Fourier transform infrared spectroscopy (FTIR) is a useful tool to study the success of GO covalent functionalization. The presence of possible crosslinkers or proteins attached to the GO material can be determined by FTIR spectroscopy based on their functional groups (e.g., Si-, COOH, and NH2 groups) and formed bonds during the functionalization (e.g., amide, ester, ether). The characteristic IR spectrum of dried GO sheets contains peaks of O-H (~ 3450 cm^-1), C=O (~ 1650 cm^-1), C-O (carboxyl group, ~ 1400 cm^-1), C-O-C (~ 1250 cm^-1), and C-OH (~

1100 cm^-1) vibrations (Figure 6).²⁹ After the protein immobilization (BSA), new peaks at 1690 cm^{- 1} (amide I; C=O, C-N), 1570 cm^-1(amide II; N-H, C-N), and 1220 cm^-1(C-N of amide group) indicate the presence of the protein and success of the functionalization.²⁹

(20)

However, IR spectra of functionalized GO materials can be challenging to interpret due to the overlapping peaks. Especially, IR vibrations of protein amide groups locate at the same region as the vibration of the imine bond (C=N, commonly formed bond between aldehydes and amines). Therefore, the separately measured reference spectra of a protein and possible crosslinkers used may be helpful.

Thermal events of materials can be studied by thermogravimetric analysis (TGA) in controlled conditions (temperature, heating/cooling rate, gas atmosphere, and pressure). In GO functionalization, TGA is used to determine the structural changes of GO material after the functionalization based on the materials’ thermal decomposition. A TG curve for GO is shown in Figure 7. Only one main thermal event can be observed: removal of labile oxygen-containing functional groups (mainly OH in the basal plane) around 200 ºC - 400 ºC. Also, in the range of 500 ºC - 700 ºC, more stable groups are removed. The thermal stability of GO is significantly lower than graphite’s, but it can improve after functionalization if labile oxygen groups are consumed during the functionalization. However, the thermal stability of BSA functionalized GO (GOS-BSA) is lower than GO, which is probably due to the incorporation of the protein.²⁹

Raman spectroscopy is commonly used to study structural defects in the graphitic lattice or oxidation degree of GO. The most important bands in the Raman spectrum of graphene and GO

(21)

are D (1350 cm^-1), G (1580 cm^{- 1}), and 2D (2690 cm^-1) bands. D band gives information about defects (sp³ hybridized carbons) in the graphene lattice. In pristine graphene, the D band is usually weak because its height is directly related to the number of the sp³ hybridized carbons.

The G band is the most intense peak in the Raman spectrum of pristine graphene, and it is originating from the in-plane vibration of the sp²hybridized carbons in the lattice. Based on the shapes, positions, and a ratio (I2D/IG) of the 2D and G bands, the number of graphene layers be concluded: For single-layer graphene, the peak shapes are sharp and symmetrical, and as the number of layers increases, the broader and at higher wavenumber the peaks are. However, the positions of the 2D and D bands are dependent on the laser excitation energy.^30,31

Raman spectroscopy can also be used to evaluate the level of functionalization of GO when functionalization results in the transformation of carbons’ hybridization (sp²to sp³) and changes in the GO’s oxidation degree. When the functionalization involves other than oxygen groups of GO (e.g., C-C double bonds of graphitic lattice), the result can be observed by Raman spectroscopy. The degree of disorder in graphene lattice can be concluded by calculating the ID/IG ratio. In the low defect density, the ratio of ID/IG increases as the disorder in graphene lattice increases. However, at higher defect density, the ratio ID/IG decreases, although the disorder further increases.^30,31 Also, an unchanged Raman spectrum after functionalization of GO could indicate the success of the functionalization, if the functionalization does not affect the properties of graphene lattice.

Figure 8 shows the characteristic Raman spectrum for GO. After the covalent immobilization of BSA on GO through diimide-activated amidation, D and G bands shift about 50 cm^-1towards lower Raman shift values. The band shifts were concluded to result from the transformation of the structure from amorphous GO sheets to partially nanocrystalline BSA-GO material. Also, during the functionalization, the number of sp²hybridized carbons slightly increased, which was confirmed by the decrease of ID/IG ratio after the functionalization (Figure 8).²⁹

(22)

Figure 8. Raman spectra of pristine GO sheet (below), and immobilized BSA on GO (above).

Atomic force microscopy (AFM) is an effective tool to analyze surface morphologies of materials. AFM gives information about the heights of particles related to the substrate’s surface.

Because GO has a flat surface morphology, except for the stacking of GO flakes, it is possible to study protein immobilization on GO surfaces.³² Average thickness of pristine GO nanosheet is around 1.1 nm (Figure 9a). After the addition of BSA, BSA molecules can be observed on top of GO nanosheets (Figure 9b).²⁹

(23)

3 GO-protein interactions

3.1 Structure of proteins

The proteins of living organisms consist of 20 amino acids, which have chemically different side chains. Amino acids can be classified into five classes based on the chemical nature of their side chains: non-polar aliphatic, polar, aromatic, positively charged, and negatively charged amino acids. Amino acids are linked together via peptide bonds, forming peptide chains of different lengths. All polypeptides have an amino terminus (N-terminus), an amino acid residue with a free α-amino group. Similarly, polypeptides have a carboxyl terminus (C-terminus) in their other end containing a free carboxyl group. When the molecular weight of a polypeptide chain is above 10 kDa, it is called protein.^33a

All proteins have a specific three-dimensional structure or structures, which are essential for their function. However, structures of proteins are not static, but they can rearrange without losing their function. A protein structure is held together mainly by weak interactions, but there can also be some covalent disulfide bonds. In a water environment, hydrophobic interactions are the main force stabilizing the protein structure: Hydrophobic side chains of amino acids are directed towards the protein interior, avoiding contact with water molecules. In addition, stabilizing hydrogen bonding and electrostatic interactions can occur.^33b

Most proteins are water-soluble because of hydrophilic amino acid residues (polar or charged) on their surfaces. However, proteins can still have some hydrophobic regions on their surfaces.

The presence of charged amino acids determines the overall charge of proteins: lysine and arginine provide a net positive charge and aspartate and glutamate a net negative charge at neutral pH. In addition to amino acids, proteins can also have additional chemical moieties bound to a polypeptide chain, such as lipids, oligosaccharides, and phosphates.^33a,b

Because of the exceptional structural and functional diversity of proteins, evaluating their interactions with GO materials is problematic. However, proteins’ surface properties can be estimated based on their size, structural stability, and chemical composition. Small and rigid

(24)

proteins (e.g., lysozyme, lactoglobulin) are structurally stable, whereas larger proteins (e.g., albumin) tend to undergo conformational changes more easily. Also, larger proteins adsorb stronger to a surface because of the larger surface contact. A protein structure can be divided into smaller domains based on the chemical nature of structural regions: hydrophilic or hydrophobic, polar or non-polar, and charged or uncharged regions.³⁴

3.2 Non-covalent interactions

GO can interact with proteins through covalent or weak interactions. Interactions involved in non-covalent protein immobilization are highly dependent on GO’s surface morphology, functional groups and oxidation degree, and surface chemistry of proteins. Weak interactions participating in protein immobilization can be hydrogen bonding, electrostatic, hydrophobic, van der Waals, and π-π interactions. Protein immobilization can retain multiple weak interactions, or one interaction can be a driving force. Especially in hydrophobic, van der Waals, and π-π interactions, surface area, electron density, and protein geometry have a key role in the formation of the interactions.^3,35

Electrostatic interactions can form between GO’s negatively charged oxygen functionalities (carboxylates and hydroxylates) and positively charged protein surfaces. The charges can also be vice versa if GO is functionalized with positively charged molecules. Because of various oxygen-containing functional groups of GO, it can act as a hydrogen bond donor or acceptor and form H-bonds with functional groups on proteins’ surfaces. On the other hand, hydrophobic graphitic regions of GO enable hydrophobic and π-π interactions with proteins. For hydrophobic interactions, the proteins must have hydrophobic amino acids on their surfaces. In addition, π-π-stacking requires that a protein contains some π-electron systems on its surface, such as an aromatic side chain of tryptophan residue.^25,35

Amino acids, which have positively charged or aromatic side chains (Figure 10) have been observed to adsorb onto GO via electrostatic interaction and π-π stacking interactions, respectively.³⁶ Order of adsorption strength was Arg > His > Lys > Trp > Tyr > Phe. Other 14

(25)

amino acids had minor adsorption capacity. Unlike the other six adsorbed amino acids, histidine was observed to adsorb on GO via both electrostatic and π-π stacking interactions. The experiments were confirmed with peptides of different amino acid sequences, showing the importance of the above-mentioned amino acids for the adsorption. Positively charged peptides containing some Lys and Arg residues attached to GO via electrostatic interactions, whereas negatively charged peptides containing some aromatic His, Trp, Phe, and Tyr residues adsorbed onto GO through π-π interactions. Additionally, negatively charged peptides containing positive Arg residues showed some adsorption on GO, but the similar peptide without Arg residues did not.³⁶ The results suggest that these amino acids (Figure 10) are the central amino acids participating in electrostatic and π-π stacking interactions in protein immobilization on GO.

Figure 10. Amino acids with positively charged or aromatic side chains at neutral pH.

Interactions between proteins and GO can be complex because of many variables. The surface charge of proteins is highly dependent on the pH and ion concentration of a buffer.²⁵ Each protein has a specific isoelectric point (pI), which is the pH in which a protein does not have any net charge. Below protein’s pI, it has a positively charged surface, and above its pI, a negatively charged surface. The further the pH of a solution is from the protein’s pI, the more charges exist on the protein surface.^33a Also, the density and chemical nature of oxygen- containing groups and the size of graphitic regions of GO can vary because of different preparation methods and storing conditions.²⁵ Therefore, there is much variation in interactions involved in protein immobilization on GO. GO-protein interactions are summarized in Table 1.

(26)

Table 1. Summary of possible weak interactions between GO and an immobilized protein

Interaction The reactive group of GO The reactive group of protein

Electrostatic Carboxylate and hydroxylate (COO^-and O^-)

Positively charged amino acids (Lys, Arg, His)

Hydrogen bonding OH, COOH and epoxy Most of the amino acids Hydrophobic Graphitic regions (sp²

hybridized carbon network)

Hydrophobic (non-polar) and aromatic amino acids (Trp, Tyr, Phe, His)

π-π stacking Graphitic regions Aromatic amino acids (Trp, Tyr, Phe, His)

3.2.1 Examples of protein immobilization

Zhang et al.³⁷ have suggested that electrostatic interactions dominate between immobilized protein (horseradish peroxidase, HRP) and GO. Protein immobilization was done by incubating HRP and GO in phosphate buffer for 30 min at 4 ºC and spontaneous attachment of the protein was confirmed by AFM. The nature of the interactions between HRP and GO was studied by repeating the experiments with phosphate buffers of different pH levels. It was observed that at acidic pH (< 7.2), HRP loading on GO was greater than at basic pH (> 7.2), which was concluded to result from attractive electrostatic interactions between positively charged HRP and negatively charged GO surface at acidic pH. It was assumed that significant electrostatic repulsion would occur above pH 7.2 because GO sheets have a total negative charge at pH ranges of 4 - 11, and HRP has a total negative charge above pH of 7.2. However, HRP loading was only 30 % greater at pH 4.8 than 8.8, indicating the presence of other interactions in the system, which were concluded to be hydrogen bonds. In addition, the biological activity of HRP reduced after immobilization owing to possible conformational changes caused by the immobilization.³⁷

Zhang et al.³⁷ also studied protein immobilization with lysozyme at pH 7, in which it has a positively charged surface. Lysozyme has a pI of 10.3, whereas HRP has 7.2, denoting that lysozyme has more positive charges on its surface around neutral pH than HRP. Protein loading of lysozyme on GO was seven times greater than the HRP loading at pH 7.0, indicating that lysozyme has more favorable surface chemistry for interaction with GO. Also, lysozyme

(27)

retained its biological activity as opposed to HRP. Zhang and coworkers³⁷ concluded that the surface charges of proteins and GO determine the interactions involved in the protein immobilization.³⁷

Usually, adsorbed proteins have a static orientation to maximize favorable interactions with the surface. However, adsorbed proteins can change their orientation if the conditions are dramatically changed. For example, at a low number of β-lactoglobulin proteins on a negatively charged surface, electrostatic interactions form between the protein and the surface. When the protein concentration increases, attractive protein-protein interactions appear, which changes the proteins’ orientation relative to the surface (Figure 11). This results in the degradation of attractive interactions between the proteins and the surface.³⁴ Therefore, in non-covalent protein immobilization, it is important to consider that the used protein concentration is favorable for protein-surface interactions.

Figure 11. Schematic representation of orientational changes of adsorbed proteins induced by electrostatic interactions. Top: Distribution of positively and negatively charged amino acids of β-lactoglobulin protein into charged domains. Middle: Protein orientation includes only protein-surface interactions at low surface density. Bottom: The number of protein-protein interactions increases at high protein surface density. Reprinted with permission from³⁴, Copyright

2011, Elsevier.

(28)

Strong adsorption of BSA on GO has been proved to occur mainly via hydrophobic interactions.³⁸ BSA (pI = 4.5) was incubated with GO in the buffer at pH 7.4. Successful BSA immobilization was confirmed by AFM and fluorescence lifetime imaging microscopy.

Interactions between immobilized protein and GO could not be electrostatic because BSA is negatively charged at pH 7.4. Therefore, the interactions were concluded to be hydrophobic, which was further confirmed by fluorescence quenching experiments. Tryptophan and tyrosine are fluorescent molecules, so hydrophobic interactions with GO can quench BSA’s fluorescence. When the concentration of GO was increased in BSA-buffer solution, the intensity of BSA fluorescence decreased.³⁸

Increasing the reduction extent of GO has been suggested to reinforce the hydrophobic interactions involved in protein immobilization.³⁹ Chemically reduced graphene oxide (CRGO) has fewer oxygen-containing functional groups than GO and resembles more pristine graphene.

Therefore, the importance of electrostatic interactions and H-bonding decreases because of a lower number of hydrophilic groups. In the study, the reduction extent of GO was controlled by the reaction time. The more GO was reduced, the higher the protein loading was. The pH of the buffer did not affect protein loading on CRGO, indicating the absence of electrostatic interactions.³⁹

Proteins used in the study (HRP and oxalate oxidase (OxOx)) are water-soluble, denoting that they have a hydrophilic surface.³⁹ Contact angle measurements, on the other hand, confirmed increasing hydrophobicity of GO with increasing reduction extent. These observations suggest that conformational changes of proteins may occur during the immobilization because the protein loading was higher on more hydrophobic CRGO than GO. In addition, the activity of HRP reduced as the reduction extent of CRGO increased, which further confirms conformational changes. However, decreasing activity was not observed for hydrophobic surface-bearing OxOx.³⁹ It is commonly accepted that water-soluble proteins can undergo conformational changes to have the more favorable orientation of hydrophobic residues for hydrophobic interactions with GO.²⁵

Dichtel and coworkers⁴⁰ have developed a method to immobilize a protein on GO by non- covalent interactions without destroying its structure by hydrophobic interactions. Unmodified GO is known to be one of the strongest inhibitors of the chymotrypsin (ChT) enzyme, which is

(29)

why ChT was used in the experiments. The method is based on the use of tripodal molecules, which have three aromatic pyrene moieties to interact non-covalently with GO’s aromatic regions. Thus, the denaturation of ChT can be prevented. Used tripod 1 has an active group that can react covalently with primary amines of proteins. ChT’s reaction with tripod 1- functionalized GO retained its enzymatic activity due to the absence of conformational changes.

ChT’s interactions with unmodified GO led to the loss of its structure and activity (Figure 12).⁴⁰

3.3 Covalent interactions

It is still unknown which amino acids of proteins participate in covalent reactions with solid substrates. In covalent GO-protein interactions, it is commonly assumed that amino and carboxylic acid groups on protein surfaces are the key groups involved in the immobilization on GO.²⁵ Figure 13 presents the most common amino acid residues used for bioconjugation of proteins with other biomolecules or chemical tags. The same residues can also be applicable for protein immobilization on GO, with or without preceding chemical modification of GO. In addition, N-terminal amino groups and C-terminal carboxylic acid groups can participate in covalent bonding, for example by forming amide linkages.^41,42

(30)

Figure 13. Schematic presentation of the distribution of commonly used amino acid residues in bioconjugation, including their average abundances and the pKa values (RNAse A

Lysine residues are the most involved in covalent reactions due to their reactive ε-amino groups and relatively high abundance on protein surfaces. Also, α-amino groups of N-terminal amino acid residues are usually on protein surfaces and hence involved in covalent reactions, but their number in proteins is much lower than lysine residues. More selective covalent binding can only be achieved by reaction only via protein’s N-terminus, especially when a protein consists of a single polypeptide chain (one reactive α-amino group).⁴¹

(31)

4 Covalent functionalization of GO

Covalent and non-covalent protein immobilization can lead to the disruption of a protein structure, which is mostly held together by non-covalent interactions, and further to protein denaturation and loss of its biological function through conformational changes. Covalent interactions usually enable more ordered and stable protein immobilization than weak interactions.⁴³ Additionally, covalent bonding enhances stability against heat, pH, solvents, and storage.³ Although non-covalent interactions usually are weaker than covalent interactions, they can rival some covalent bonds when they occur over a large surface.³⁵

Protein immobilization on a solid support via weak interactions results in random protein orientation. Although covalent protein immobilization can result in specific protein orientation on a support, it requires that both a protein and a support have specific groups which can react together selectively. In the case of a protein, it means that a specific chemical group is added to a certain position of a protein so that the protein conformation and function are retained. Some of the covalent protein immobilization protocols are based on the use of natural, unmodified proteins, which can more often result in the loss of protein conformation after protein-GO interactions than the use of modified proteins.⁴⁴ This chapter presents both random (crosslinking) and site-specific approaches (click reactions, Staudinger ligation) for covalent protein immobilization on GO.

4.1 Diimide-activated amidation

Diimide-activated amidation (Figure 14) is a commonly used and relatively straightforward method for the covalent attachment of proteins to materials. It has also been used for covalent protein immobilization on graphitic materials such as carbon nanotubes and GO.⁴⁵ The amidation reaction occurs between activated carboxyl groups of GO and amino groups of proteins.²⁹ Proteins with many lysine or arginine residues on their surfaces (e.g. bovine serum albumin) are probably favorable for the amidation due to the greater number of reactive amino

(32)

groups. The main disadvantage of the method is that GO may contain a low number of COOH groups, so the functionalization will remain low.⁴⁶

Figure 14. A general reaction path of diimide-activated amidation.

Protein functionalization of GO via diimide-activated amidation has three main steps (Figure 15). In the first step, COOH groups of GO are activated with N-ethyl-N'-(3-dimethyl aminopropyl) carbodiimide (EDAC) hydrochloride forming reactive O-acylisourea intermediate. Then, N-hydroxysuccinimide (NHS) is added forming succinimidyl intermediate (step 2), which reacts with the amino groups of protein, forming an amide bond between chemically modified GO and a protein (step 3).²⁹ NHS is needed in the reaction becauseO- acylisourea intermediate is very reactive in an aqueous environment, and its lifetime may not be long enough for the reaction with a protein. Without NHS, intermolecular conjugation of proteins rich in amino and COOH groups on their surfaces can occur.⁴⁵

(33)

4.2 Carboxylation and amination

GO usually contains fewer COOH groups than other functionalities, and they usually locate at the edges of GO sheets. However, more COOH groups on the basal plane of GO can be synthesized by carboxylation of epoxy and OH groups, which enhances GO’s water solubility.

A widely used method is to treat GO with chloroacetic acid 2 under strongly basic conditions such as sodium hydroxide-water solution (Figure 16). Chloroacetic acid 2 reacts with epoxy groups through a ring-opening reaction forming OH and O-COOH groups. OH groups can further react to O-COOH.⁴⁷

Figure 16. General carboxylation protocol of GO with chloroacetic 2 acid and sodium hydroxide.⁴⁷

Faghihi et al.⁴⁷ have studied the efficiency of the GO carboxylation method. GO carboxylation was performed in four different chloroacetic acid concentrations (0.5, 1, 2, and 3 M) using a 4 M NaOH water solution. Quantitative amounts of COOH groups of GO and carboxylated GO samples were estimated by methylene blue (MB) assay, as an MB molecule can interact covalently with a COOH group. Hence, the change in MB concentration before and after the reaction can estimate the number of COOH groups in GO. Based on the experiments, the number of COOH groups increased significantly after carboxylation with 1 M or 2 M chloroacetic acid.⁴⁷

A large amount of COOH on the GO surface may, however, be unfavorable for covalent protein immobilization: The GO surface can be highly negatively charged, which attracts positively charged molecules via electrostatic interactions. Also, the folding of GO sheets can be more probable when the number of COOH is substantial.⁴⁷

(34)

Carboxylation of GO requires strongly basic conditions. However, strong bases, such as NaOH with chloroacetic acid, can cause partial reduction of GO resulting in a lower number of oxygen- containing functional groups. Bianco et al.⁴⁸ have observed partial reduction of GO during carboxylation by the same method mentioned above. C/O ratio was higher after carboxylation, and the number of COOH groups increased, but the number of C-O bonds decreased, indicating the removal of labile oxygen-containing groups.⁴⁸

Bianco et al.⁴⁸ also studied how efficient GO carboxylation is in the GO double functionalization (Figure 17). At first, GO was functionalized with aminated PEG 3 through an epoxide ring-opening reaction. Then, epoxy and OH groups of the amine-functionalized GO were carboxylated, leading to a significantly lower number of amine groups than before carboxylation. After diimide-activated amidation of COOH groups with an aminated PEG polymer chain 3, the number of amine groups increased but was still remarkably lower than after the first NH2-PEG functionalization. However, a more efficient NH2-PEG double functionalization of GO was achieved without carboxylation, although the number of GO COOH groups was lower.⁴⁸

Royal Society of Chemistry, https://creativecommons.org/licenses/by-nc/3.0/.

(35)

Instead of carboxylation, GO can be aminated for further functionalization or direct reaction with the COOH group of a protein. Epoxy groups of GO are reactive to nucleophiles and will react through ring-opening reactions. Salvio and coworkers⁴⁹ have developed a method to synthesize amino-functionalized GO flakes. An epoxide ring-opening reaction was conducted using sodium azide, which led to the formation of OH and azide groups. Then, azide groups were reduced to amines using the reducing agent lithium aluminum hydride (LiAlH4). Although the method resulted in a high degree of functionalization, its total duration only for GO functionalization is almost one month, and the reaction can also form explosive intermediates.⁴⁹

4.3 Crosslinking

4.3.1 Glutaraldehyde

Proteins can be attached to GO via chemical crosslinking, which means that a protein and GO are connected via a crosslinker molecule. In polymer chemistry, crosslinking means covalent linkages between the polymer chains. This idea can also be applied to the covalent protein immobilization on GO. By using crosslinkers or spacer molecules to attach a protein to GO, its quaternary structure and hence biological function can be retained. Crosslinkers increase the distance between GO and protein when direct absorption via non-covalent interactions on GO reduces.²⁵

A widely used crosslinking agent for bioconjugation is glutaraldehyde 4 (GA; Figure 18). It has also been used for covalent protein immobilization on GO. Advantages of using GA are good solubility in water and organic solvents and high reactivity. However, in aqueous solutions, GA can exist in various forms, such as monomer, dimer, or polymer chains, depending on pH and other chemical species in the solution (Figure 18).⁵⁰ Approximately 13 possible structural forms of GA in aqueous solution complicate the prediction of its reaction outcomes. Also, polymerization causes a problem, as it diminishes the organization of the functionalization.

However, polymerization can be avoided by using small amounts of GA relative to the quantity

(36)

of GO.^51,52 GA polymer chains can be converted back to monomers at neutral or basic pH by heating or sonication.⁵⁰

Figure 18. Some possible structural forms of GA in the aquatic environment at different pH.⁵⁰

There is still no agreement on the primary structure of GA in each pH, but many suggestions have been presented. In some studies, commercially available GA solutions have been observed to be mainly mixtures of GA polymers, such as forms 5-7 at neutral or alkaline pH (Figure 18).

Also, forms 8 and 9 have been suggested to exist in alkaline solutions. Only a tiny amount of form 5 was found at acidic pH, and GA was predicted to be mainly in form 4 and its hydrated forms 10, 12, and 13 (equilibrium between the forms). On the other hand, it has been proposed that acidic aqueous solutions of GA could contain structures of 4, 10, and 11.⁵⁰ Because of many suggested GA structures in an aqueous solution, GA is presented as its monomer 4 in the GO functionalization reactions for clarity.

Aldehydes react with amines forming imines and with alcohols forming hemiacetals and acetals, depending on the stoichiometric amount of alcohol.⁵³ Hence, in GO functionalization, GA will react with hydroxyl groups of GO forming hemiacetal or acetal bridges (assuming that GA is in form 4).^51,52 The order of addition of GO or GA may determine how aldehyde groups of GA react with hydroxyl groups of GO. Tan et al.⁵¹ have shown that when GA is slowly (dropwise)

(37)

added to the GO-water dispersion, GA forms crosslinks between GO sheets via both terminal aldehydes (Figure 19a). However, if the GO dispersion is added slowly to the GA solution, GA molecules react only via one terminal aldehyde group forming GA grafted graphene oxide nanosheets (Figure 19b). This is most likely due to the smaller number of available hydroxyl groups.⁵¹

After the covalent functionalization of GO, structural changes can be semiquantitatively analyzed with the unvarying C=C peak near 1630 cm^-1. The FTIR spectra of GA crosslinked GO (GA-GO2) and GA grafted GO (GA-GO) show differences: The higher intensity of C=O stretching at ~1724 cm^-1 (aldehyde) compared with C=C peak indicate the unreacted aldehyde groups in GA-GO compared with GA-GO2 (Figure 20). The corresponding relative intensities between GA-GO2 and non-functionalized GO were close to each other (0.61 and 0.59), indicating no remarkable change in the number of aldehyde functionalities after GA functionalization.⁵¹

(38)

Successful covalent binding of GA to GO can be observed from several FTIR peaks (Figure 20). The peaks at 2800-3000 cm^-1correspond probably to the C-H stretching of GA. Also, the relative intensities of C=C (~1625 cm^-1) and C-O (alkoxy, ~ 1100 cm^-1) peaks increase after GA binding, which indicates the increase in the number of alkoxy groups and suggests the presence of covalent bonds between GA and GO.⁵¹ It has also been observed that the relative intensities of OH-group peaks at 3300-3500 cm^-1 reduced remarkably after GA or glyoxal binding to GO. This indicates that these dialdehydes react with OH groups of GO.⁵⁴

As some aldehyde groups of GA remain unreacted after the reaction with GO, they can further react with NH2 groups of proteins, such as a free amino group in the N-terminus of the polypeptide chain forming imine bonds (Schiff bases, enhanced in alkaline pH).⁵⁵ It has been experimentally proved that GA is most reactive towards unprotonated ε-amino groups of lysine amino acids.⁵⁶ Although most lysine residues are protonated at acidic and neutral pH, a low number of unprotonated lysine residues are sufficient to transfer acid-base equilibrium towards deprotonation. As polar moieties, lysine residues usually locate on protein surfaces. Lysine side chains are generally not in enzymes’ active sites, which means that the enzyme’s conformation and biological activity retain after modification to a lysine residue.⁵⁰ Histidine and tyrosine residues can also react with GA, but they are less reactive than free amino groups.⁵⁶

(39)

Figure 21 presents the proposed reactions of monomeric and polymeric forms of GA with an amino group of protein. The most common way to describe the reaction is Schiff base formation, in which imine bond forms between amino and aldehyde groups (Figure 21a). The formed Schiff base is very unstable, especially in acidic pH, and will easily return to free starting materials. However, Schiff base formation is enhanced in alkaline pH, and imine bond can be converted to a more stable imine linkage by using a reducing agent.^57a In addition to a Schiff base formation, many other reactions may happen due to various structures of GA. Figure 21b presents possible reaction pathways of polymeric GA and protein molecules. In reaction 1, a protein reacts with GA’s aldehyde group, forming a Schiff base, stabilized by conjugation in the GA polymer chain. Another possibility is reaction 2, where a protein reacts with an ethylenic double bond. However, this reaction requires an excess of amino groups.⁵⁰

Figure 21. Proposed reactions between GA and a protein. a) Schiff base formation.^57a b) Schiff base formation stabilized by conjugation (reaction 1), and conjugated addition of a

protein to the ethylenic double bond of GA polymer (reaction 2).⁵⁰

Lactoperoxidase enzyme (LPO) has been covalently immobilized on GO flakes via GA crosslinker (pH 6.8),⁵⁵ which can also crosslink GO sheets together (Figure 22). The thermal stability of crosslinked LPO was better than free LPO’s stability, and crosslinked LPO had maximum enzymatic activity at a higher temperature and more alkaline pH than free LPO.

Formed bonds between LPO’s lysine residues and GA resulted in a net anionic charge to the system, which might cause the pH change of LPO. The increase in optimum pH, on the other hand, can result from the reduced free movement of crosslinked LPO.⁵⁵

(40)

4.3.2 Silane crosslinkers

The surface of GO can be functionalized with organosilanes and further other crosslinkers. In the presence of water, silanes hydrolyze easily into reactive silanols, and they can form Si-O- Si bonds on the surface of GO through silanol condensation (Figure 23a).^57b However, the formation of silane coating requires that silanols are close to each other. For example, surface silanization of GO have been done with (3-aminopropyl) triethoxysilane (APTES) under mild reaction conditions.⁵⁸ APTES can react either via its amino group forming amine linkage with GO’s epoxy group or via its ethoxysilane group forming Si-O-C bonds with OH groups of GO (Figure 23b). Serodre et al.⁵⁸ have proposed that both reactions occur simultaneously, resulting in amino and silanol groups on the surface of functionalized GO. The number of epoxy and OH groups decreased at the same time as amine and Si-O-C bonds formed, which was confirmed with XPS, FTIR, and TGA-FTIR measurements.⁵⁸

(41)

Figure 23. a) Reaction of GO with APTES and formation of silane coating through silanol condensation in water. ^57b b) Three possible types of reactions between APTES and GO in the

presence of water.⁵⁸

Laccase enzyme has been immobilized on GO via a double crosslinker system, 3-aminopropyl trimethoxy silane (APTMS) and GA.⁵⁹ APTMS reacts with GO similarly to APTES. However, reaction with GO’s OH groups is more favorable because it leaves free amino groups for GA crosslinking.⁵⁸ After amine functionalization, GA is added to insert aldehyde groups to GO, which can further react with amino groups of protein, such as laccase enzyme (Figure 24).⁵⁹

Figure 24. Schematic representation of laccase immobilization on GO using APTMS and GA crosslinkers.⁵⁹

(42)

4.4 Click reactions

Sharpless et al.⁶⁰introduced the term “click chemistry” in 2001 to describe fast, “spring-loaded”, and highly selective organic reactions. In a typical click reaction, covalent bond forms between carbon and a heteroatom. The reaction has a high thermodynamic driving force (> 20 kcal/mol), enabling fast completion of the reaction. Also, click chemistry reactions have a high yield of final products, easy purification, high stereospecificity, and simple reaction conditions.⁶⁰ In biomedical applications, four click reaction types are commonly used; cycloaddition, nucleophilic ring-opening, non-aldol carbonylation, and carbon-carbon multiple bond addition reactions.⁶¹ However, cycloaddition reactions are mainly utilized in functionalization of GO.

4.4.1 Copper-catalyzed azide-alkyne cycloaddition (CuAAC)

The most typical click reaction utilized in GO functionalization is copper-catalyzed cycloaddition of azides and alkynes (CuAAC), which forms stable triazole linkage in acidic or basic organic solvents. Copper (I) catalyst is needed in the reaction temperatures favorable for GO and proteins. Functional groups participating in azide-alkyne cycloaddition are unreactive toward biological molecules, such as proteins. Hence, the reaction is highly specific. Azide- alkyne reactions have been used for protein conjugation, so it would also be applicable for covalent protein immobilization on GO.^57c

Alkynyl terminated poly(ethylene)glycol (PEG) and amino acids have been connected to azide- functionalized GO sheets utilizing the cycloaddition approach (Figure 25). First, COOH groups of GO were activated with EDC and NHS as in diimide-activated amidation. After that, 3- azidopropan-1-amine 14 reacted with the modified GO by EDC condensation, resulting in azide-functionalized GO. Then, alkynyl terminated amino acid or PEG chain was added, and selective cycloaddition of alkyne and azide took place (Figure 25).⁶² This approach can be applicable also for larger macromolecules such as proteins and enzymes because they can be modified to have an alkyne group.⁴⁴

(43)

Figure 25. Covalent functionalization of GO with azide and addition of amino acids and polymers through CuAAC reaction. Adapted from the article⁶².

Successful covalent attachment of azide groups on GO was confirmed from the FTIR spectrum, where the characteristic peak of azide groups at 2098 cm^-1can be observed. After the click reaction, the azide peak disappeared, and the peak at 1100 cm^-1indicated the stretching of C-O groups of PEG.⁶²

The cycloaddition can also occur the other way round, between alkynyl functionalized GO and azide functionalized peptide (Figure 26). In Shi and coworkers’ study,⁶³ hydroxyl groups of GO were converted to alkynyl groups through silanization with alkynyl-containing silane coupling agent 15. The CuAAC reaction of functionalized GO and the azide-functionalized peptide was then performed.⁶³ The peptides used in the experiments were RGD peptides, a class of peptides enhancing cell adhesion. The abbreviation RGD signifies that this sort of peptides always contain an amino acid sequence of arginine (R), glycine (G), and aspartic acid (D).⁶⁴

Covalent functionalization of graphene oxide with proteins